| { | |
| "checkpoint_path": "qwen2vl-evaluator", | |
| "train_json": "distillation/data/format_gpt_data_balanced_order.json", | |
| "eval_json": "", | |
| "output_dir": "distillation/checkpoints/qwen2vl-evaluator-sft-balanced-order", | |
| "epochs": 3, | |
| "batch_size": 1, | |
| "grad_accum_steps": 8, | |
| "lr": 1e-05, | |
| "weight_decay": 0.0, | |
| "warmup_ratio": 0.03, | |
| "max_length": 3072, | |
| "max_grad_norm": 1.0, | |
| "seed": 42, | |
| "fp16": false, | |
| "bf16": true, | |
| "logging_steps": 10, | |
| "save_steps": 0, | |
| "clear_cache_steps": 50, | |
| "max_train_samples": 0, | |
| "max_eval_samples": 0, | |
| "max_image_size": 512, | |
| "wandb_project": "qwen2vl-evaluator-sft", | |
| "disable_wandb": true | |
| } |